Compute Core, Cures and Robots

The idea

I believe the AI buildout runs through a handful of bottlenecks, not just the famous chip names, so the core here is the foundry, equipment makers, networking, and power and cooling gear alongside the well known chip designers. Around that I hold smaller, steadier satellites in large pharma, medtech, industrial automation, and surgical robotics, meant as anchors rather than lottery tickets. I keep a small cash buffer and avoid concentrating in any single name, planning to hold this for years unless the bottleneck thesis itself breaks down.